Character Recognition using an artificial neural network. Network is trained for 26 characters. Implemented in C++ and Magick++ API is used for pixel manipulation.
Network is build with 4 layers, Input layer, Output layer and 2 hidden layers with 80 and 40 nodes. Used Sigmoid and Tangent-Sigmoid functions for the node Activation.
Network is trained for 26 English uppercase characters from A-Z using 6 training cases for each distinct character.
| Input |  |
 |
 |
 |
 |
|---|---|---|---|---|---|
| Output | Recognized: C |
Recognized: W |
Recognized: S |
Recognized: A |
Recognized: D |